Conferencing solutions are increasingly moving to digital implementations based on servers, rather than bridging circuits or other analog connection techniques. In such server-based environments, the quality of the conferencing experience for conferees may depend on how heavily the server (and its central processor or CPU) is loaded at a given time. If the server becomes overloaded, the conferencing experience may deteriorate, and may eventually become unacceptable.
In light of the above considerations, previous approaches have tracked metrics, such as CPU utilization, to determine how heavily the server is loaded. However, CPU utilization may not provide a completely accurate metric of how heavily the server is loaded. Moreover it may not have a linear inverse relationship to the conference experience, making it hard to judge the quality of experience.